Simulating Fundamental Tax Reform in the U.S.

by

David AltigFederal Reserve Bank of Cleveland

Alan J. Auerbach University of California, Berkeley and NBER

Laurence J. KotlikoffBoston University and NBER

Kent A. Smetters University of Pennsylvania

Jan WalliserInternational Monetary Fund

revised, September 29, 1999

The views expressed here are those of the authors and do not necessarily reflect those of the FederalReserve Bank of Cleveland, the IMF, or any other organization. This paper was written while JanWalliser was an analyst at the Congressional Budget Office. We are grateful to Cristina DeNardi,Dennis Epple, Barbara Fried, Bill Gale, Jane Gravelle, participants in workshops at the Universityof Chicago, the NBER, the Federal Reserve Bank of Cleveland, Indiana University, and the FederalReserve System Committee on Macroeconomics, and two anonymous referees for comments onearlier drafts. Auerbach thanks the Burch Center for Tax Policy and Public Finance and Kotlikoffthanks the National Institute of Aging and the Smith Richardson Foundation for research support.

Abstract

This paper uses a new, large-scale, dynamic life-cycle simulation model to compare the welfare andmacroeconomic effects of five fundamental alternatives to the U.S. federal income tax: aproportional income tax, a proportional consumption tax, a flat tax, a flat tax with transition relief,and a progressive variant of the flat tax called the X tax.

The model incorporates intragenerational heterogeneity and detailed specification of alternative taxsystems. It predicts significant long-run increases in output from the wholesale replacement of theprogressive federal income-tax with a low-rate proportional consumption tax. In the long run,middle- and upper-income groups gain from this policy. But their gains come at the expense of thecurrent and future poor, who suffer from the removal of the current tax structure’s progressivity, andthe wealthiest members of initial, older generations, who are hurt by the policy’s implicit capitallevy.

The flat tax adds a standard deduction to the consumption tax. This feature improves outcomes forthe current and future poor. But it does so at an additional cost; long-run output increases less thanhalf as much as under the proportional consumption tax, and middle-income groups experience awelfare loss, even in the long run. Moreover, the flat tax continues to leave some of the initial elderlyworse off. Although one can insulate these individuals through transition relief, doing so furtherreduces the long-run gains from tax reform.

Switching to a proportional income tax without deductions or exemptions is an alternative way toreform the federal income tax. Although the policy raises long-run output to roughly the same extentas the flat tax, eliminating the federal income tax’s progressivity hurts the poor, whether they areinitially alive or are born after the reform. The X tax, a flat tax with a progressive subsidy to wages,makes everyone better off in the long run and also raises long-run output by more than the standardflat tax. But its higher tax rate on existing capital hits initial older generations particularly hard.

Nearly all of the reforms analyzed would be approved by a majority of the existing adult population,based on the impact on their individual well-being. However, the degree of approval bears littlerelationship to a policy’s long-run benefits.

JEL: H20, C68

1Auerbach and Slemrod (1997) argue that the problems of the 1981 tax reform prompted passage of the TaxReform Act of 1986.

2A similar model, used to consider only steady states, is presented in Altig and Carlstrom (forthcoming).

I. Introduction

Fundamental tax reform has been a hot issue, and for good reason. The U.S. tax system —

a hybrid of income- and consumption-tax provisions — is complex, distortionary, and replete with

tax preferences. Recent “reforms” of the tax code, including the Taxpayer Relief Act of 1997, have

made the system even more complex and buttressed the argument for fundamental reform.1

“Fundamental tax reform” means different things to different people. The definition adopted

below is the simplification and integration of the tax code by eliminating tax preferences and taxing

all sources of capital income at the same rate. Several current tax proposals certainly deserve to be

called “fundamental.” They include Hall and Rabushka’s (1983, 1995) flat tax, the retail sales tax,

and Bradford’s (1986) X tax. The flat tax and the retail sales tax are two alternative ways of taxing

consumption. The X tax also taxes consumption, but places high-wage earners in higher tax brackets

than low-wage earners. Another fundamental reform is to adopt a broad-based, low-rate income tax.

This paper uses a computable general equilibrium simulation model to compare the welfare

and macroeconomic effects of fundamental tax reform. The model is a substantially enhanced

version of the Auerbach-Kotlikoff (1987) dynamic life-cycle simulation model.2 The new model

follows the significant lead of Fullerton and Rogers (1993) in incorporating intra- as well as

intergenerational inequality. Specifically, it includes 12 different groups within each cohort, each

with its own earnings ability (its own endowment of human capital).

Our new model approximates U.S. fiscal institutions much more closely than does its

predecessor. It includes an array of tax preferences, a progressive social security system, and a

3The Fullerton-Rogers model, in contrast, assumes that all agents face the same marginal tax rate independentof income.

4See CBO (1997).

2

Medicare system. Including tax preferences is not only crucial for studying fundament tax reform.

It also permits our use of actual tax schedules in calibrating the model.3 Stated differently, omitting

tax preferences would mean unrealistically low tax rates since the current federal income tax covers

only 57 percent of national income.4 The improved modeling of social security and Medicare is also

vital since both programs materially alter the intergenerational and intragenerational distributions

of welfare and the initial set of fiscal distortions from which tax reform proceeds.

Like Auerbach and Kotlikoff (1987), we compute the economy’s perfect foresight transition

path. Given the magnitude of factor-price and tax-rate changes along our simulated transition paths,

permitting agents to think rationally about the future is of great importance. This and other

advantages vis a vis the Fullerton-Rogers (1993) model, which assumes myopic expectations, must

be set against some disadvantages. Our model has a simpler production and preference structure

than does the Fullerton-Rogers model, which features multiple consumption and capital goods,

intermediate inputs, and industry-specific capital income taxation. As such, it cannot measure the

efficiency gains from the removal of inter-sectoral and inter-asset distortions. Nor does it

incorporate the externalities present in endogenous growth models, which might provide another

source of efficiency gains from tax reform (e.g. Stokey and Rebelo 1995).

Our model also omits the impact on labor supply of low-income programs such as the Earned

Income Tax Credit, and several influences on saving, including earnings and lifespan uncertainty,

transfer program asset tests, and liquidity constraints. The impact of these factors has been evaluated

5To be precise, in each tax reform simulation the levels of government purchases and outstanding debt are heldconstant through time when measured in effective units of labor.

3

in simulation studies by Hubbard and Judd (1987), Hubbard, Skinner, and Zeldes (1995), and Engen

and Gale (1996). The low intertemporal elasticity of substitution used in our simulations is, in part,

a reflection of the fact that not all saving is driven by standard life-cycle concerns. However, the

lack of a richer model should be borne in mind in assessing our findings.

We use our model to examine five fundamental tax reforms that span the major proposals

currently under discussion. Each reform we consider replaces the federal personal and corporate

income taxes in a revenue-neutral manner.5 The reforms are a proportional income tax, a

proportional consumption tax, a flat tax, a flat tax with transition relief, and the X tax.

The proportional income tax applies a single tax rate to all labor and capital income, with no

exemptions or deductions. The proportional consumption tax differs from the proportional income

tax by permitting 100 percent expensing of new investment. One may think of it as being

implemented via a wage tax at the household level plus a business cash-flow tax. The flat tax differs

from the proportional consumption tax by including a standard deduction against wage income and

by exempting implicit rental income accruing from the ownership of housing and consumer durables.

The remaining two proposals modify the flat tax to address distributional concerns. The flat tax with

transition relief aids existing asset holders by permitting continued depreciation of old capital

(capital in existence at the commencement of the reform). The X tax aids lower-income taxpayers

by substituting the flat tax’s single-rate wage tax with a progressive wage tax. To recoup the lost

revenue, its sets the business cash-flow tax rate equal to the highest tax rate applied to wage income.

Alternatively, one can think of the X tax as a high rate flat tax with a progressive subsidy to wages.

4

Each of the reforms broadens the tax base, permitting reductions in marginal tax rates on

labor supply and saving. So too does the implicit tax on existing wealth that arises under the four

proposals that shift the tax structure toward consumption taxation by introducing full expensing.

The expensing of new capital effectively eliminates the taxation of capital income at the margin.

However, unlike the simple elimination of capital income taxes, this tax benefit is available to new

capital only, and, consequently, reduces the value of existing capital relative to that of new capital

in a manner equivalent to that of a one-time tax on their wealth. As discussed in Auerbach and

Kotlikoff (1987), this capital levy is crucial to both the efficiency and long-run welfare gains from

switching to consumption taxation. It permits a permanent reduction in distortionary marginal tax

rates and shifts the burden of paying for government spending from young and future generations

to middle-age and older initial wealth owners.

As indicated, the five reforms differ in the treatment of marginal and inframarginal capital

income, the extent of base-broadening, and progressivity. These differences translate into different

income and substitution effects on consumption and labor supply which, in turn, generate the

different responses to the five policies. These responses depend on our choices of parameter values

and our assumptions about capital adjustment costs. In what may be our most realistic results — in

particular, including capital adjustment costs — we find the following:

The proportional income tax raises the long-run level of output by almost 5 percent. It also

generates sizable increases in the capital stock and the supply of labor. However, the reform hurts

the poor, who face low effective rates of income taxation under the current federal income tax system

due to its deductions and exemptions. The proportional consumption tax raises long-run output by

over 9 percent. The implicit wealth tax generated by this reform reduces the welfare of the initial

5

middle aged and elderly. They respond to this and the increase in after-tax interest rates by

consuming less and working more, which raises national saving and investment. The expanded

capital stock and reduced fiscal burden make most of those alive in the long run significantly better

off. However, eliminating tax progressivity lowers the welfare of the poorest members of society.

The flat tax’s standard deduction alleviates some of the distributional concerns raised by the

proportional income and proportional consumption taxes. But this deduction increases the tax rate

needed to satisfy the government’s intertemporal budget constraint. Consequently, long-run output

rises by less than 5 percent. Although the flat tax’s standard deduction insulates the poor from

welfare losses, it hurts middle-income groups, especially in the short run but even in the long run.

Its capital levy also hurts initial high-income elderly cohorts. Those welfare losses must be set

against the welfare gains enjoyed by all groups in the long run.

Adding transition relief to the flat tax limits the welfare losses of initial capital owners. But

this modification of the flat tax reduces aggregate income gains again, with long-run output now

rising by less than 2 percent. Furthermore, because replacement tax rates must increase to

compensate for the lost revenue associated with transition relief, all but the richest and poorest

lifetime-income groups suffer welfare losses in the long run. The X tax, which raises long-term

output by 6.4 percent, provides no transition relief. It also confronts the rich with higher effective

tax rates on their labor supply. It is not surprising, then, that the X tax helps those who are poor in

the long run by more than it helps those who are rich. Still, all long-run cohorts gain.

Thus, fundamental reform of the U.S. tax system offers significant long-run economic gains,

but these gains come at the expense of certain groups. Modifications that mitigate adverse transition

and distributional effects also substantially reduce the long-run gains. In considering the plausibility

6

of the results, one should bear in mind that these changes are generally much more radical than any

U.S. fiscal policy change enacted in recent memory. Moreover, the structure of economic incentives

and the distribution of resources are being altered in two ways — by the policies themselves and by

their general equilibrium impacts on the time paths of factor prices. Hence, simple intuition about

the expected magnitude of individual responses is difficult to apply to the reforms being considered.

In a final exercise, we consider the extent to which any of the reforms considered might be

approved by a majority of the population alive at the time of enactment. Although this analysis

should not be viewed as an explanation of the political determinants of tax reform, it is instructive

nonetheless. It reveals two interesting results. First, a majority of the existing population gains in

all cases considered. Second, the share of the existing population that gains is not a good measure

of a policy’s long-run benefits. Even one policy that reduces long-run output would improve the

welfare of around two-thirds of the existing population.

II. The Model

Demographic Structure

The model’s agents differ by their dates of birth and their lifetime labor-productivity

endowments. Every cohort includes 12 lifetime-earnings groups, each with its own endowment of

human capital and pattern of growth in this endowment over its lifetime. The lifetime-earnings

groups also differ with respect to their bequest preferences. All agents live for 55 periods with

certainty (corresponding to adult ages 21 through 75), and each j-type generation is 1+n times larger

than its predecessor. At age 21, each j-type cohort gives birth to a cohort of the same type.

Population growth is exogenous, and each cohort is (1+n)20 larger than its parent cohort.

7

Preferences and Household Budget Constraints

Each j-type agent who begins her economic life at date t chooses perfect-foresight

consumption paths (c), leisure paths (l), and intergenerational transfers (b) to maximize a

time-separable utility function of the form

(1) U jt '

1

1& 1[ j75

s'21s&21 c j

1&1

s,t%s&21 % l j1&

1

s,t%s&21

1&1

1&1

% 54µj b j1&

1

75,t%54 ].

In (1), is the utility weight on leisure, is the intertemporal elasticity of substitution in the

leisure/consumption composite, and is the intratemporal elasticity of substitution between

consumption and leisure. The parameter µj is a j-type specific utility weight placed on bequests left

to each child when the agent dies, representing a “joy of giving” bequest motive. The term =

1/(1+ ), where is the rate of time preference, is assumed to be the same for all agents.

The relationship between , , and the elasticity of labor supply with respect to the current

wage is spelled out in the first section of the Appendix. There are many different labor supply

elasticity concepts found in the literature. However, for our purposes of relating empirical estimates

to the parameters of a life-cycle utility function, the most useful concept is the “Frisch” labor supply

elasticity, corresponding to variation along an optimal path, with marginal utility of income constant.

Letting as,tj be capital holdings for type j agents, of age s, at time t, maximization of (1) is

subject to a lifetime budget constraint defined by the sequence:

a js%1,t%1 ' (1%rt) (a j

s,t%g js,t) % w j

s,t(Ej

s,t&l js,t) & c j

s,t & jKk'1 T k(B j,k

s,t ) & Nb js,t,

6Both of these adjustments, to the slope of the wage profile and to the time endowment, are needed toincorporate the impact of technical progress on wage growth, the former because we assume that each cohort’s timeendowment is fixed at birth. See Auerbach et al (1989) for a detailed discussion of this approach.

8

(2)

l js,t # E j

s,t,

and

a j75,t $ 0,

where rt is the pretax return to savings, gs,tj are gifts received from parents, Es,t

j is the time

endowment, bs,tj are bequests made to each of the N = (1+n)20 children, and the functions Tk(@) (with

tax bases Bs,tj,k as arguments) determine net tax payments from income sources k=1,...,K. All taxes

are collected at the household level, and the tax system includes both a personal income tax and a

business profits tax. There are no liquidity constraints, so the assets in (2) can be negative, although

terminal wealth — the wealth left over after final period bequests are made — must be non negative.

An individual’s earnings ability is an exogenous function of age, type, and the level of

labor-augmenting technical progress, which grows at a constant rate . We summarize all skill

differences by age and type via an efficiency parameter sj. Thus, the wage rate for an agent of type

j and age s is ws,tj = s

j wt, where wt is the real wage at time t. The term sj increases with age to

reflect not only the accumulation of human capital, but also the technical progress that occurs over

the course of each individual’s life; i.e., the values of sj are set to establish a realistic longitudinal

age-wage profile. To permit balanced growth without a further restriction of preferences (i.e., to

keep the ratio of labor supply to labor endowment constant in the steady state), we model the growth

in lifetime wages from one generation to the next as growth in time endowment rather than in the

wage rate per unit of time; that is, we assume that technical progress causes the time endowment of

each successive generation to grow at rate .6 Thus, if Es,tj is the endowment of type j at age s and

9

time t, then Es,tj = (1+ ) Es,t-1

j, for all s, t, and j. Because E grows at rate from one cohort to the

next, technical progress imparts no underlying trend to wt.

Transfers are received by children, with interest, at the beginning of the period after they are

made by their parents. We restrict all parental transfers to bequests, so that bs,tj = 0, for s ú 75, and

gs,tj = 0, for s ú 56. In the steady state, therefore, gj = bj, for all j (where we have dropped age

subscripts for convenience).

The Government

At each time t, the government collects tax revenues and issues debt (Dt+1) that it uses to

finance government purchases of goods and services (Gt) and interest payments on the inherited

stock of debt (Dt). Letting j stand for the fraction of j-type agents in each generation, the

government’s official debt evolves according to:

(3) Dt%1 % (1%n)t j12j'1

j j75s'21 (1%n)&(s&21) jK

k'1 T k(B j,ks,t ) ' Gt % (1%rt)Dt.

Government purchases are assumed to be either a) unproductive and generate no utility to

households or b) be fixed and enter household utility functions in a separable fashion. The values

of Gt and Dt are held fixed per effective worker throughout the transition path. Any reduction in

government outlays resulting from a change in the government’s real interest payments is passed on

to households in the form of a lower tax rate.

The model also has a social security system that incorporates Old-Age and Survivors

Insurance (OASI), Disability Insurance (DI), and Medicare’s Hospital Insurance (HI). Old-age

benefits are calculated according to the progressive statutory bend-point formula. Disability and

7See chapter 10 of Auerbach and Kotlikoff (1987) for a more detailed discussion.

10

Medicare benefits are provided as lump-sum transfers. The OASI payroll tax is set at 9.9 percent

and applied to wage income up to a limit of $62,700. HI and DI tax rates are set at 2.9 percent and

1.9 percent, respectively. Like the OASI tax, DI contributions apply only to wages below $62,700.

The HI tax, by contrast, is not subject to an earnings ceiling.

Benefits are scaled to reflect spousal and survivor benefits using distributional information

provided in the 1997 OASDI Trustees Report. We set the perceived marginal link between the

OASI contributions and the OASI benefits at 25 percent. The perceived effective OASI tax rate is,

thus, 7.4 percent—75 percent of 9.9 percent.7 Lump-sum HI and DI benefits are provided on an

equal basis to agents above and below age 65, respectively.

Firms and Technology

Aggregate capital (K) and labor (L) equal the respective sums of individual asset and labor

supplies as indicated in equations (4) and (5).

(4) Kt ' (1%n)t j12j'1

j j75s'21 (1%n)&(s&21) a j

s,t & Dt,

(5) Lt ' (1%n)t j12j'1

j j75s'21 (1%n)&(s&21) j

s (E js,t&l j

s,t).

Output (net of depreciation) is produced by identical competitive firms using a neoclassical,

constant-returns-to-scale production technology. In the base case, the aggregate production

technology is the standard Cobb-Douglas form:

11

(6) Yt ' AKt L 1&t ,

where Yt is aggregate output (national income) and is capital’s share in production. However, in

later simulations, we do consider the impact of a lower elasticity of substitution in production, ,

using a constant-elasticity of substitution (CES) function.

Another important aspect of the production technology in our base case is the assumption that

it is costly to adjust the capital stock. Tax reforms, particularly those that eliminate the marginal tax

on capital income, can induce large increases in demand for capital. The extent to which increased

demand translates into more investment, rather than higher asset values, affects both the efficiency

of the tax reform and its distributional consequences. We model adjustment costs as convex, indeed

as a simple quadratic function of investment:

(7) C(It) ' 1 % 0.5 It /Kt It

with the quadratic term equal to 10, a value consistent with the recent results of Cummins et al.

(1994). With convex adjustment costs, investors will be induced to smooth out increases in

investment, and the shadow price of new capital goods relative to their replacement cost — Tobin’s

q — will move in the same direction as investment.

The competitive pre-tax, pre-expensing rate of return to capital at time t is given by the

marginal product of capital (defined in terms of the capital-labor ratio, )

(8) Rt ' A &1t .

8This equals Rt, as defined in (8), plus the reduction in proportional adjustment costs due the increase in thecapital stock. See Auerbach (1989) for a derivation of equations (9) and (10).

12

In general, tax systems treat new and existing capital differently. Under the consumption tax,

new capital is permitted immediate expensing, while existing capital receives no such deduction.

Even under the existing income tax, the combined effect of accelerated depreciation and the lack of

inflation-indexing makes the depreciation allowances per unit of existing capital lower than those

given new capital. We model provisions that treat new and existing capital differently using the

mechanism of fractional expensing of new capital, at rate z. That is, we set z to account for the

extent to which new capital faces a lower effective tax rate than does existing capital (with z=1 under

the consumption tax). If tK is the time-t marginal tax rate on net capital income (i.e., the tax rate

applied to capital income net of expensing) then, given (7), arbitrage between new and existing

capital implies that the latter has a unit value of

(9) qt ' 1&ztKt % (1& K

t ) It / Kt ,

assuming that adjustment costs are expensed. Equation (9) equals Tobin’s q. Note that q depends

not only on the strength of demand for new capital (via the second term), but also on the relative

treatment of old and new capital. Tax reforms typically affect both of these terms.

The arbitrage condition arising from profit-maximization implies that the post-tax return is:

(10) r t 'Rt % 0.5 It / Kt

2 (1& Kt ) % qt%1 & qt

qt

.

In (10), the total return to capital includes its after-tax marginal product8 plus capital gains.

13

III. Calibration

Much of our model’s parameterization is relatively standard. Exceptions include

earnings-ability profiles and the fiscal structure. We turn first to these elements and then discuss

more familiar preference and technology parameters. Table 1 summarizes our selected parameters.

Earnings-Ability Profiles

The growth-adjusted earnings ability profiles in equation (5) are of the form

(11)js ' e

j0 % ( %

j1)s %

j2s 2

%

j3s 3

where, as discussed above, is the constant rate of technical progress. Values of the coefficients

for j-type groups 1 through 12—in ascending order of lifetime income—are based on regressions

fitted to the University of Michigan’s Panel Study of Income Dynamics, using a strategy similar to

that in Fullerton and Rogers (1993). The procedure involves (i) regressing the log of hourly wages

on fixed-effect dummies, cubics in age, and interactions between age, age-squared, and a set of

demographic variables; (ii) using the estimated coefficients from step (i) to generate predicted

lifetime wage profiles; (iii) sorting the data according to the present-value of implied lifetime

income, and dividing the sorted data into the 12 classes according to lifetime-wage income; and (iv)

estimating the coefficients of (11) from the simulated data profiles of each of the 12 groups.

In sorting the data for steps (iii) and (iv), we divided the population into deciles. Groups 1

and 12 comprise the bottom and top 2 percent of lifetime wage income earners, and groups 2 and 11

the remaining 8 percent of the top and bottom deciles. Each other group constitutes 10 percent of

9As the payroll tax in our model is used entirely to pay for current benefits, we set the rate slightly below its 15.3percent statutory rate to account for the portion of payroll taxes going toward trust-fund accumulation. See Table 2.

14

the population. For example, group 3 is the second decile of lifetime-wage income, group four the

third decile, and so on up to group 10. The second section of the Appendix describes this procedure.

Figure 1 presents the estimated earnings-ability profiles, scaled to include the effects of

technical progress. Given our benchmark parameterization, peak hourly wages valued in 1996

dollars are $4.00, $14.70, and $79.50 for individuals in classes 1, 6, and 12, respectively. More

generally, steady-state annual labor incomes derived from the model’s assumptions and from the

endogenous labor supply choices range from $9,000 to $130,000. As discussed below, these

calculations include labor compensation in the form of fringe benefits.

Fiscal Structure

The model includes government purchases of goods and services, government debt, and

distortionary taxes. The level of government purchases, Gt, was chosen so that the benchmark

steady-state ratio of government purchases to national income equals 0.211. The level of government

debt, Dt, was chosen such that the associated real interest payments equal about 3.1 percent of

national income in the initial steady state. These values match the corresponding 1996 values for

combined local, state, and federal government in the United States.

The benchmark tax system in our initial steady state is designed to approximate the salient

aspects of the 1996 U.S. (federal, state, and local) tax and transfer system. It features a hybrid tax

system (incorporating wage-income, capital-income, and consumption tax elements) and payroll

taxation for the social security and Medicare programs.9 To adjust for tax evasion, we reduce income

taxes by 2.6 percent. This adjustment is consistent with the degree of tax evasion reported in

10We use a quadratic approximation rather than the exact, discrete brackets to simplify the simulation problem.Using a differentiable tax function allows us to derive first-order conditions that do not involve kink points and shadowwages at these kink points. This simplification is particularly relevant because our household decision problem alreadyhas a kink point at the minimum taxable income level, and a potential corner solution at zero hours worked, both of whichinvolve the calculation of shadow wage rates, as well as a nonconvexity at the point where the social security earningsceiling is reached. These additional complications are discussed below in the section that discusses the solution of themodel.

11The data used in this estimation were taken from all taxable returns in tax year 1993. The function wasobtained by regressing deductions exclusive of mortgage interest expense on the midpoints of reported income ranges.(The deduction of interest expense on home mortgages was included in our calculation of the capital-income tax rate,as we will subsequently describe.) The regression yielded a coefficient of 0.0755 with an R2 equal to 0.99.

12Benefits as a function of adjusted gross income were kindly provided by Jane Gravelle of the CongressionalResearch Service and Judy Xanthopoulos of the Joint Committee on Taxation. Based on this information we regressedtotal benefits on AGI. The regression yielded a coefficient of 0.11295 with an R2 equal to 0.99. In defining the wage-taxbase, we therefore exempt roughly 11 percent of labor compensation from the base calculations.

15

Slemrod and Bakija (1996). In the alternative tax structure experiments we assume that evasion

reduces the post-reform tax base (income net of deductions and exemptions) by the same percentage

as before the reform. This is a simplifying assumption that merits exploration in future research.

We approximate the current U.S. tax system by specifying a progressive wage-income tax,

a flat capital-income tax, a flat state income tax, and a flat consumption tax.

Wage Income Taxation

The wage-income tax structure has four elements: 1) a progressive marginal rate structure

derived from a quadratic approximation to the 1996 federal statutory tax rates for individuals10; 2)

a standard deduction of $4000 and exemptions of $5660 (which assumes 1.2 children per agent,

consistent with the model’s population growth assumption); 3) itemized deductions, applied only

when they exceed the amount of the standard deduction, that are a positive linear function of income

estimated from data reported in the Statistics of Income11; and 4) earnings-ability profiles that are

scaled up to incorporate pension and fringe components of labor compensation.12

16

Capital Income Taxation

Since our model has a single capital good, we need to calibrate marginal and inframarginal

capital income tax rates based on the average values of these rates taken across different types of

U.S. capital. Here, we follow closely the calculations in Auerbach (1996), who reports that income

from residential capital and non-residential capital are taxed at flat rates of 6 percent and 26 percent,

respectively. Given that the U.S. capital stock is split evenly between these two forms of capital, the

weighted average federal marginal tax rate on total capital income is about 16 percent. Because of

the difference in treatment of new and existing capital, primarily in the nonresidential sector, U.S.

capital faces a higher tax rate — roughly 20 percent for the capital stock as a whole, according to

Auerbach’s estimates. To incorporate these tax rates in our model, we assume that all capital income

faces a 20 percent tax, but that 20 percent of new capital is expensed, thereby generating a 16 percent

effective rate on new capital. In addition to the federal taxation, both capital and wage income are

subject to a proportional state income tax of 3.7 percent.

Consumption Taxation

Consumption taxes in the initial steady state reflect two elements of the existing tax structure.

First we impose an 8.8 percent tax on consumption expenditures, consistent with National Income

and Product Account values for indirect business and excise revenues, a substantial component of

which are state sales taxes. However, because contributions to both defined benefit and defined

contribution pension plans receive consumption tax treatment, we add an additional 2.5 percent tax

on household consumption expenditures to account for the indirect taxation of labor compensation

17

in the form of pension benefits (Auerbach 1996). This 2.5 percent tax replaces the wage tax that

otherwise would apply if pension contributions were taxed as income.

Preferences and Technology

Our initial choices for the remaining technology, preference, and demographic parameters

are summarized in Table 1. The value for , about which there is little evidence, is set equal to .004

to generate a realistic value for the capital-output ratio in the initial steady state. The values of , ,

and n are those in Auerbach and Kotlikoff (1987). The intertemporal elasticity, , is set equal to .25,

representing a relatively low degree of substitution between consumption at different dates. While

some studies in the literature have found higher elasticities, most have not. One possible reason is

the presence of liquidity constraints and other factors that may mitigate the responsiveness of saving

to interest rate changes. Our use of a low intertemporal elasticity of substitution serves as an

imperfect proxy for such factors.

The parameter is chosen such that agents devote, on average, about 40 percent of their

available time endowment (of 16 hours per day) to labor during their prime working years (roughly

ages 21-55). Note that the choice of also affects the implied elasticity of labor supply. As we

discuss in the Appendix, the combined values of , , and generate Frisch labor supply elasticities

that are well within the range of empirical estimates. Still, as this range is fairly large, we also report

results for values of and that induce smaller behavioral responses. The parameters µj are chosen

to match bequests as a fraction of income in the initial steady state based on estimates by Menchik

and David (1982) reported in Fullerton and Rogers (1993).

18

Solving the Model

Following Auerbach and Kotlikoff (1987), we solve the model with a Gauss-Seidel

algorithm. The calculation starts with guesses for certain key variables and then iterates on those

variables until a convergence criterion is met. The model’s identifying restrictions are used to

compute the remaining economic variables as well as the updates for the iterations. The solution

involves several steps and inner loops that solve for household-level variables before moving to an

outer loop that solves for the aggregate variables including the time-paths of capital stock and

aggregate labor supply.

The household optimization problem is subject to the constraint that leisure not exceed the

endowment of time (equation 2). For those households who would violate the constraint, the model

calculates shadow wage rates at which they exactly consume their full-time endowment. The

household’s budget constraint is kinked due to the tax deductions applied against wage income under

the personal income tax. A household with wage income below the deduction level faces marginal

and average tax rates equal to zero. A household with wage income above the deduction level faces

positive marginal and average tax rates. Due to the discontinuity of the marginal tax rates, it may

be optimal for some households to locate exactly at the kink. Our algorithm deals with this problem

as follows. We identify a household that chooses to locate at the kink in particular periods by

evaluating each period’s leisure choice and corresponding wage income above and below the kink.

We then calculate a set of period-specific shadow marginal tax rates from the period-specific

first-order conditions that put such households exactly at kinks in each period in which being at a

kink is optimal. This calculation of shadow tax rates for particular periods is simultaneous; i.e., a

shadow tax rate in any particular period will influence labor supply decisions in all other periods.

19

The payroll tax ceiling introduces additional complexity by creating a non-convexity in the

budget constraint. For those above the payroll tax ceiling, the marginal payroll tax rate on labor

earnings is zero. For each period, we evaluate the utility on both sides of the non-convex section and

put households on the side that generates highest utility.

Aggregate variables of the model are solved with a forward-looking algorithm that iterates

on the capital stock and labor supply over the entire transition path. Initial guesses are made for a)

the time-path of aggregate demands for capital and labor, b) each household’s shadow wage and tax

rates at each age, and c) the endogenous tax rate (for which the program is solving), the payroll tax

rate, and social security and Medicare benefit levels. Given the initial guesses of the time-paths of

all these variables, the model calculates a) the factor prices in each period that are consistent with

the use by firms of the aggregate inputs assumed to be demanded and b) the remaining lifetime

consumption and leisure choices for all income classes in each current and future cohort. Shadow

wages and shadow taxes are calculated to ensure that the time endowment and the tax constraints

discussed above are satisfied. Households’ labor supply and assets are then aggregated across

cohorts and, within cohort, across lifetime income classes for each period. This aggregation

generates a new guess for the time-paths of the aggregate supplies of capital stock and labor supply.

In equilibrium, the factor supply time paths for capital and labor must equal their

corresponding factor demand time paths. Hence, to form a new guess of the time paths of aggregate

factor demands we form weighted averages of the initial guess and the supply time paths derived

using the previous guess of the time path of factor demands. The time paths of the tax rate for which

we are solving and the payroll tax rate are also updated to meet the revenue-neutrality requirement

13Note that the social security replacement rate and absolute level of Medicare benefits are exogenous.

14This number is somewhat lower than the estimated 1996 percent return to capital relative to replacement costof 9.3 percent listed in Table 2. In our model with nonzero adjustment costs, the before-tax interest rate determined byexpressions (9) and (10) differs from the return to capital, because market value differs slightly from replacement cost(see (9)) and because the total return to capital accounted for by the interest rate also includes the reduction in adjustmentcosts brought about by additions to capital (see the numerator in (10)). Ignoring these two terms, the return to capitalitself, equal to the term R in (8), is 9.7 percent.

20

and to preserve the pay-as-you-go financing of social security and Medicare benefits.13 The

algorithm then iterates in this manner until the capital stock and labor supply time-paths converge;

i.e., until the time paths of factors demanded are consistent with the time paths of factors supplied.

The Benchmark Equilibrium

Table 2 provides summary statistics for the initial steady state. Given our parameter choices,

the model generates a pre-tax, pre-expensing interest rate of 8.3 percent14, a net national saving rate

of 5.1 percent, and a capital/national-income ratio of 2.6. Consumption accounts for 73.1 percent of

national income, net investment for 5.1 percent, and government purchases of goods and services

for 21.1 percent. These figures are close to their respective 1996 NIPA values.

The calibrated model’s initial economy-wide average marginal tax rate on wage income is

21.6 percent, close to the figure obtained from the NBER’s TAXSIM model reported in Auerbach

(1996). The average wage-income tax rate equals 12.2 percent. For all individuals in the highest

lifetime income class (group 12), the average effective marginal tax rate on labor income is 29.2

percent. The highest realized effective marginal tax rate is 34.9 percent. For lifetime income class

6—whose members have peak labor earnings of about $35,000—the average tax rate and average

15The average marginal rate for people with the lowest income exceeds zero due to positive shadow tax ratesin peak earnings years.

21

marginal tax rate are 10.7 and 20.0 percent, respectively. For the poorest class (group 1), the

corresponding rates are zero and 11.1 percent.15

In this initial steady state, bequest wealth (the accumulated value of inheritances received

summed over all households) depends on the age structure of bequest receipt. Our assumption that

individuals receive bequests at age 56 is made primarily for simplicity, as there is scant evidence on

the actual age structure of bequest receipt. With this assumption, and our calibration of bequests

themselves to the data cited earlier, bequest wealth represents 30 percent of the capital stock. This

percentage is smaller than that reported for 1974 for overall transfer wealth by Kotlikoff and

Summers (1981). The percentage would, presumably, be larger had we assumed an earlier age of

bequest receipt, but still is in rough agreement with the Kotlikoff-Summers findings concerning the

amount of wealth generated solely by bequests as opposed to intervivos transfers, which we do not

model explicitly. That said, the importance of bequests and intervivos transfers for U.S. capital

formation appears to have diminished through time because of the remarkable increase in recent

decades in the degree of annuitization of the elderly (see Auerbach et al 1995).

IV. Initial Tax Reform Simulations

Table 3 summarizes the five tax-reforms. Table 4 presents simulations for all five reforms

for the base case assumptions. In each of these simulations, our initial steady state is the same,

calibrated to the 1996 U.S. economy as described above. Table 4 also presents variables of interest

for this initial steady state and for three transition years, 1997, 2010, and 2145, meant to illustrate

22

short-run, medium run and long-run effects. Subsequent tables, described in the next section, present

the results of sensitivity analysis involving preference and technology parameters.

A Proportional Income Tax

Our first experiment replaces the progressive tax on wage income and the proportional tax

on capital income with a proportional tax applied to all income. In addition, the proportional income

tax eliminates the major preferences in the federal income tax, including the standard deduction,

personal and dependent exemptions, itemized deductions, and the preferential tax treatment of fringe

benefits. The last of these is implemented by decreasing the consumption tax rate by 0.025 and

subjecting all compensation to the new proportional income tax. The investment expensing rate

remains at its initial 20 percent level.

The first panel in Table 4 summarizes the aggregate results from this reform. The marginal

tax rates required to satisfy the government’s budget constraint stay close to 13 percent over the

entire transition path. This value lies far below both the 21.6 percent average marginal rate applied

to labor income and the 16 percent rate applied to capital income in the benchmark steady state.

National income rises by 3.8 percent immediately and by 4.9 percent ultimately. In the early years

of the transition, these output changes are dominated by increased work effort associated with the

lower marginal tax rates. In the long run, higher wealth levels mitigate some of the increase in labor

supply. However, the accumulated effects on the stock of capital from the reform more than

compensate for the reduced labor supply: in the long run the capital stock increases by 5.6 percent.

The short-run decrease in the capital-labor ratio produces a short-run increase in the before-tax

interest rate and a short-run decrease in before-tax wage rate. The long-run increase in the

16We focus on the welfare effects of tax reforms on members of different income groups and generations, ratherthan on the overall efficiency change that might be calculated by aggregating such individual welfare effects. As ourmodel’s disaggregation by income class is largely secondary in the context of an efficiency analysis, it seems unnecessaryto repeat the extensive analysis of efficiency already provided in Auerbach (1996), based on the original AK model.

17For members of the fifty-five transition generations, an individual’s age at the time of reform equals the 21minus the number on the horizontal axis.

23

capital-labor ratio produces the opposite effects on factor prices. This reform’s initial impact on the

stock market (measured via Tobin’s q) is positive, reflecting the increase in saving and demand for

capital that results from two factors: the rise in short-run disposable income (due to higher labor

supply) and, to a lesser extent (because it doesn’t change much), the lower marginal tax rate on

capital income. The second potential effect on q, associated with changes in the relative treatment

of old and new capital, is small because the level of expensing has not changed and the effective tax

rate on capital income has decreased only slightly (housing capital is no longer exempt).

Figure 2 shows the effects of the tax reform on remaining lifetime utility for different

generations by lifetime-income group.16 For ease of exposition, the figure reports the utility gains

only for classes 1, 3, 6, 9 and 12. The horizontal axis of the figure lists the period the generation

enters the model — reaches adulthood — relative to the period of the regime shift (period 0). For

example, -1 refers to the generation that reaches adulthood just prior to the regime shift, 0 to the

generation that reaches adulthood in the period of the shift, 1 in the following period, and so on.17

The change in remaining lifetime utility is measured as the equivalent variation of remaining full

lifetime income. In interpreting these numbers, one should keep in mind that full lifetime income

includes the value of leisure. In our model, full lifetime income is more than twice the size of

remaining actual lifetime earnings. Hence, gains or losses will tend to be larger if measured relative

to either realized earnings or consumption.

24

In the long run, only members of lifetime-income groups 8 through 12 benefit from the

proportional income-tax reform, the rise in aggregate output notwithstanding. The main reason is

that average tax rates increase for income classes 1 through 7 due to the loss of deductions and

exemptions. In the short run, however, the oldest agents at the time of the reform benefit slightly

since the reform increases the after-tax return to capital.

A Proportional Consumption Tax

Our proportional consumption tax differs from the proportional income tax by including full

expensing of investment expenditures. The government is now taxing income less domestic

investment, which, in our closed economy, equals income less saving, i.e., consumption. Formally,

we specify the consumption tax as a combination of a labor-income tax and a business cash-flow tax.

The second panel of Table 4 summarizes aggregate effects. The first thing to note is the drop in q

— the value of the existing capital stock relative to new capital. This drop occurs even though the

rate of investment surges, because the impact of investment demand on q is more than offset by the

sharp increase in the tax advantage of new versus old capital. This second effect constitutes the one-

time effective tax on existing capital assets mentioned above.

In the period just after the tax reform, labor supply increases by 6.3 percent, a higher effect

than was observed under the proportional income tax. This is because there are now additional

factors at work beyond the reduction in marginal labor income tax rates. The rise in after-tax interest

rates (because capital income is now essentially untaxed) produces substitution effects that

encourage delays not only in consumption, but also in leisure, impacts reinforced by the negative

18There is also an income effect associated with the rise in interest rates, equal to the reduction in the presentvalue of future consumption less the present value of future labor income. These income effects will generally bepositive and discourage saving, though the effect will be much smaller than in the simple two-period life-cycle modelwith first-period labor supply, in which only the future-consumption term is present.

25

wealth effect among those holding old capital.18 These two factors also generate a substantial short-

run jump in the saving rate from 5.1 percent to 7.3 percent. However, as the initial negative wealth

effects diminish over time and interest rates fall with the growth in capital, the saving rate eventually

recedes to 5.9 percent. Fourteen years into the reform (in 2010), the capital stock (per effective unit

of labor)is 10.8 percent larger than its initial steady state value, and output is 6.3 percent larger. In

the long run, the capital stock exceeds its initial value by 25.4 percent, and output exceeds its initial

value by 9.4 percent, strong reactions that permit the consumption tax rate to fall over time, from

14.2 percent initially to a long-run 12.7 percent.

Figure 3 shows that, despite the large aggregate income gains, lower lifetime-income groups

are hurt by the reform. Although these losses are not as large as those in the proportional income-tax

case — indeed, several groups switch from being long-run utility losers to long-run utility winners

— the regressive nature of the outcomes persists. The figure also reveals welfare losses for initial

rich elderly, who own the lion’s share of the existing capital stock. In contrast, the initial poorest

elderly gain from the tax reform. There are two reasons. First, this group consumes almost entirely

out of social security benefits, the real value of which are unaffected by the change in asset values.

Second, this group also borrows against some of their social security benefits prior to retirement, and

their slightly negative net worth in old age shrinks in magnitude due to the policy-induced fall in the

value of existing capital.

19To impose the capital levy on existing housing would require taxing current owners’ imputed rent and eventualsale proceeds.

26

The Flat Tax

Our flat tax experiment modifies the proportional consumption tax by including a standard

deduction of $9500. In addition, it fully exempts housing wealth — about half of the capital stock

— from taxation. Because policy makers are unlikely to extend the capital levy to housing, this

exemption is an important step toward realism.19

As the third panel in Table 4 makes clear, the need to finance the standard deduction and tax

exemption of existing housing increases the replacement tax rates well above those of the

proportional consumption tax. As a result, the output effects under the flat tax are substantially

reduced relative to its proportional counterpart. The long-run rise in the capital stock and level of

output are, respectively, only 59 and 48 percent as large as those under the proportional consumption

tax. The labor supply response is lower as well, reflecting the higher short- and long-run levels of

marginal tax rates. The revenue-neutral flat-tax rate equals 21.4 percent initially and reaches 19.9

percent in the long run. The short-run impact on Tobin’s q is quite similar to that of the proportional

consumption tax, but this masks two offsetting effects. Investment demand rises less under the flat

tax, moderating the positive impact of this factor on q. On the other hand, as the flat tax exempts

much of existing wealth from the implicit tax old capital, it also moderates this factor’s negative

impact on q. In the longer run, as the first difference declines in importance (with net investment

converging to zero in both instances), the second effect predominates, resulting in a higher value of

q under the flat tax.

27

Figure 4 shows that the flat tax generates short-run utility effects that are similar to those of

the consumption tax. The long-run utility changes that are bunched much more closely than those

of either of the proportional tax reforms. This reflects the flat tax’s attention to preserving

progressivity. Interestingly, the highest relative gains are for the richest and poorest lifetime-income

groups, with the poorest gaining from the higher standard deduction and the richest from the

flattened marginal rate schedule.

The utility changes for the richest and poorest lifetime-income groups also differ from those

of the middle groups throughout the entire transition path. Group 12 benefits most from reduced

marginal and average tax rates. Group 1, which pays very little taxes under either regime, benefits

from the overall increase in wages. For those in income groups 3 through 9, the marginal and

average tax rates initially change little or even rise. This stems from the revenue neutrality of the

experiment, which requires a flat-tax marginal rate that initially exceeds the pre-reform tax rates for

some agents in the middle-income classes in order to finance the lower tax rates at the top end.

Those who belong to the lifetime middle-income range and enter the workforce close to the time of

reform suffer utility losses along the transition path. They face relatively high tax rates of 20 to 22

percent on labor income for 20 to 25 years of their working life before the growth of the capital stock

becomes fully effective. Once the economy grows, though, tax rates fall and wages rise, which

leaves group 3 better off and raises the lifetime utility levels of groups 6 through 9 nearly to the point

of indifference.

Neither the macroeconomic variables nor the welfare effects of the flat tax experiment are

substantially influenced by the existence of a bequest motive. We repeated the simulation by

“turning off” bequests (setting the utility bequest weight µj = 0 for all j), simultaneously reducing the

20Because of the extreme similarity to the simulation just presented, the results for this experiment are not shownseparately in the table.

28

rate of time preference slightly (from .004 to .002) to maintain the same initial steady-state capital-

output ratio. The resulting transition path was nearly identical to that just discussed, with the long-

run increases in national income, the capital stock, and labor supply equal to 4.6 (versus 4.5) percent,

15.3 (versus 15.0) percent, and 1.4 (versus 1.3) percent, respectively, and no significant differences

in the remaining initial steady-state computations nor in the post-reform changes in output or

welfare.20 While it might seem surprising that eliminating bequests has so little impact, it should be

remembered that bequests are modeled here as basically another type of future consumption. Thus,

the motivations for bequest saving and life-cycle saving are essentially the same.

The Flat Tax with Transition Relief

One important characteristic of consumption tax reform is its treatment of existing assets and

their owners. In contrast the positive welfare gains they receive under the switch to proportional

income taxation, older, higher-wealth asset holders lose with the adoption of the proportional

consumption tax or the flat tax. Though these losses are moderated by the surge in asset demand that

limits the decline in existing asset values, the presence of such losses may could make transition

relief for existing assets a political necessity. Our fourth experiment adds transition relief to the flat

tax by extending pre-reform depreciation rules for capital in place at the time of the tax reform.

Since the present value of depreciation allowances equals roughly 50 percent of the nonresidential

capital stock, transition relief is modeled by cutting the effective cash-flow tax rate in half.

As the fourth panel of Table 4 confirms, all of the salutary long-run aggregate effects of the

standard flat tax are mitigated by the introduction of transition relief, which must be financed by

29

permanently higher tax rates. The transition relief is, in some sense, excessive, in that the value of

existing assets actually rises slightly in the short run. Still, the capital stock increases by over 8

percent in the long run, affording a 1.9 percent rise in the long-run level of output. Labor supply

changes little following this tax reform, actually declining slightly below its initial steady-state level,

reflecting both higher marginal tax rates (on average) and positive wealth effects.

Figure 5, which shows the welfare effects of the flat tax with transition relief, differs

markedly from Figure 4, which shows the effects with no transition relief. Transition relief replaces

the small short-run welfare losses of the wealthier initial elderly with sizable welfare gains, and

raises the welfare gains of all income classes who are in their middle ages when the reform is begun.

But these welfare gains come at the cost of smaller welfare gains for certain future generations and

welfare losses for others. For example, with no transition relief, the long-run rich (members of

groups 12 alive in the long run) experience more than a 1.6 percent gain in welfare. But with

transition relief, these gains are cut nearly to zero. For middle-class households alive in the long run,

the concession to initial wealth holders transforms the flat tax from having essentially no impact to

a roughly 1.5 percent loser when measured in terms of its welfare impact. Indeed, with the exception

of the very poorest and very richest members of society, transition relief transforms the flat tax into

a bad deal for those alive in the medium and long runs. This is critically important to keep in mind

in assessing calls for a flat tax. Many advocates of the flat tax favor transition relief, apparently

without sufficient understanding that offsetting wealth effects are already present, and that short-run

relief undermines the results they claim the flat tax will achieve in the long run.

21Recall that marginal wage-tax rates are a linear function of taxable labor income. The adjustments requiredto maintain budget balance are implemented by changing the intercept of this function, while holding the slope constant.

30

The X Tax

The X tax, using the present-law standard deduction, maintains the progressivity of the

present-law wage tax schedule. It also sets the cash-flow tax rate equal to the highest marginal tax

rate on labor income, in this case 30 percent.21 An important reason for choosing equal rates of tax

on labor and cash flow under the flat tax was that doing so eliminates the incentive for businesses

to shift income. Under the X tax, the cash flow tax rate cannot equal all marginal tax rates on labor

income simultaneously, but setting it equal to the highest labor-income tax rate is nearly as effective,

as most potential shifting would involve high-income executives and business owners in the top

labor-income bracket. However, an interesting byproduct of this design feature is that it raises the

levy on old capital above that imposed by the basic flat tax. This higher capital levy offsets the rise

in marginal tax rates on labor income due to progressivity, for it reduces the present value of revenue

that labor income taxes must raise.

As shown in the bottom panel of Table 4, this reform produces large long-run output gains

despite maintaining progressive wage taxation. Only the proportional consumption tax generates

larger long-run output gains. The long-run welfare effects of the X tax are progressive. As shown

in Figure 6, the long-run gains vary inversely with lifetime income. However, the percentage gains

across groups are closer for this proposal than for any other, suggesting that, in the long run, it is —

as intended — the least disruptive in its distributional impact. All groups experience welfare gains

of between 1 and 2 percent of full lifetime resources. In the short run, though, those in the highest

income class who are old at the time of the reform suffer the largest welfare loss — reaching 2

31

percent of remaining lifetime resources — since they hold the largest share of physical assets. The

poorest elderly, on the other hand, actually benefit from the capital levy since they live essentially

on their social security benefits which are, in fact, a source of a slight amount of borrowing.

V. Sensitivity Analysis

This section considers the impact of changes in assumptions regarding four key parameters,

for two of the tax reforms just considered, the standard flat tax and the flat tax with transition relief..

(In each instance, we make offsetting parameter adjustments to maintain the correspondence between

our initial steady state and the benchmark economy.) Table 5 presents the results of reductions in

two production parameters, the adjustment cost term and the elasticity of substitution in production

between labor and capital, . Table 6 presents the results of reductions in the two elasticities in the

household utility function, the intratemporal elasticity and the intertemporal elasticity .

Production Parameters

Our base case assumes that there are convex costs to adjusting the capital stock. When

investment increases, these costs lead to a smoothing of capital accumulation, and a higher level of

initial asset values when investment surges. These changes, in turn, should each have a negative

effect on short-run labor supply, as slower capital accumulation moderates the increase in real wages,

and higher wealth encourages the consumption of leisure. At the same time, the higher initial asset

values serve to diminish the impact of the tax on existing wealth, encouraging consumption by the

elderly and reducing capital accumulation and longer-run output growth.

Although the literature has implicated adjustment costs as a cause of sluggish investment

responses, it has proven more difficult to uncover the asset revaluation effects with which such

22We alter the value of the time preference parameter, , to .015, in order to maintain the same initial capital-output ratio. This is the same discount rate used in Auerbach and Kotlikoff (1987) for the no-adjustment-cost case.

32

adjustment costs should be associated (e.g., Cutler 1988). Thus, it is useful to consider the effect of

omitting adjustment costs and their associated effects on valuation. This is done in the top panel of

Table 5 which repeats the two flat tax simulations with the adjustment cost term, , set to zero.22

These simulations bear out the intuition just given. While, in the long run, labor supply

effects are quite similar in the presence and absence of adjustment costs, eliminating adjustment

costs makes the short-run labor supply responses much larger. In the standard case, labor supply

rises immediately by 3.0 percent, rather than by 1.6 percent. With transition relief, labor supply rises

slightly, rather than falling. The other important short-run difference associated with adjustment

costs is the behavior of q, which is lower, as expected, in both cases. In the longer run, output is

higher than in the absence of adjustment costs, due to the higher amount of capital accumulation.

Figures 7 and 8 show the effects on welfare under these two simulations. The long-run

patterns are similar to those in Figures 4 and 5, which correspond to the same two tax reforms with

adjustment costs present, although the welfare gains are higher without adjustment costs, reflecting

the higher level of capital accumulation. However, the short-run welfare patterns are quite different.

In particular, higher-wealth elderly fare far worse when adjustment costs are left out, because of the

significant decline in their asset values. This comparison suggests that the existence of adjustment

costs, while reducing the long-run gains from tax reform, also substantially mitigates the short-run

transition losses that such reforms generate.

The second panel of Table 5 presents simulations for the CES production function,

23As we reduce the production elasticity, we also change the capital intensity parameter, , from 0.25 to 0.312,and multiply the production efficiency parameter, A, by 0.962. It is a matter of simple algebra to show that these twochanges ensure that the levels of capital and labor and the wage and interest rates in the initial steady state are the sameas for the base case with Cobb-Douglas production.

33

(12) Yt ' A[ K 1&1/t %(1& )L 1&1/

t ]1

1&1/

in which the production elasticity of substitution, , is reduced from 1 (i.e., Cobb-Douglas) to 0.8,

a reasonable alternative value to our base case assumption.23 As one would expect, the lower

elasticity of substitution means that the before-tax interest rate falls by more as capital accumulates,

somewhat dampening the capital accumulation incentive. Thus, while the long-run interest rate is

8.0 percent in the base case for the standard flat tax displayed in Table 4, it is 7.8 percent in the

corresponding run in Table 5 for = .8; the long-run gain in the capital stock is reduced from 15.0

percent to 13.6 percent. However, these differences, as well as those in other variables, is small.

The same conclusion holds for the case of transition relief.

Preference Parameters

As discussed above, there is considerable uncertainty about the “correct” values of the

parameters and . It is therefore important to determine how sensitive the results presented thus

far are to reasonable variations in these parameters. In each case, because we pose the question in

terms of whether tax reform might fail to produce welfare and income gains, we consider smaller

values of these elasticity parameters, which reduce the degree to which agents react to improved

incentives. However, one can use these simulations to infer the effects of raising or . The top

panel of Table 6 presents results for each of the two flat tax variants, with the value of the

24We adjust , to -.08 to preserve the initial capital-output ratio when changing . However, when changing

, we adjust the labor intensity parameter , to 0.2, in order also to maintain the appropriate share of labor supply in totallabor endowment.

34

intratemporal elasticity of substitution, , reduced from .8 to .4. The bottom panel covers the same

two reforms, this time with the intertemporal elasticity of substitution, , reduced from .25 to .1.24

The experiment with a smaller intratemporal elasticity of substitution looks quite similar to

the flat-tax case under our benchmark parameterization (in Table 4) in the early stages of the

transition path. However, in the long run, the decreased willingness to substitute higher consumption

for leisure results in a decline in aggregate labor supply relative to the higher elasticity case. The

smaller increase in labor supply contributes to a smaller increase in the level of saving (even though

the saving rate is similar) and capital accumulation. As a result, capital growth is smaller as well

and, although real wage growth is similar to before, the slower growth in both labor and capital

makes income growth smaller as well, only 2.8 percent in the long run, instead of 4.5 percent. The

relative impact is similar when transition relief is added.

The effects of the smaller intertemporal elasticity of substitution are greater, because this

change reduces the responsiveness not only of labor supply, but of consumption as well. Thus, in

addition to the smaller labor supply response just observed, we also see a smaller increase in the

saving rate. As a result, short-run and long-run output gains are even smaller than in the upper panel

of the table. Under the pure flat tax, output rises by just 1.3 percent in the long run; with transition

relief added, long-run output actually falls by 3.1 percent. Thus, there is no guarantee that a realistic

consumption tax reform — the flat tax with transition relief — will raise output in the long run.

35

VII. Voting for Tax Reform

While the previous reforms vary in their short-run and long-run consequences, any

consideration of reform must confront a very basic question: if the reform appears to be welfare-

improving, why hasn’t it already happened? One answer, of course, is that our calculations ignore

some important economic factor that makes reform less attractive. Another, though, could be that

reforms that provide long run benefits to most individuals, such as the X tax, do not necessarily help

those alive at the time of reform.

Describing the nature of political equilibrium is beyond the scope of this paper. However,

it is interesting to address a simpler question, whether the reforms would appear to be supported by

a majority of the existing population. Recall that, although we have incorporated a bequest motive

in the utility function, it is not the type for which members of initial generations would incorporate

the welfare of those in future generations. With a “joy of giving” bequest motive, individuals are

not intergenerationally altruistic. Hence, given the chance, they will vote for policies that improve

their welfare at the cost of reducing that of future generations.

Table 7 presents, for each simulation in Tables 4, 5, and 6, the fraction of the population alive

at the commencement of reform that gains from the policy change. Perhaps the most surprising

aspect of the table is that, in all cases, a strong majority of the existing population favors reform.

The largest share of the population favors the X tax. Looking at Figure 6, we may interpret this as

the outcome of two factors. First, the X tax imposes a considerable tax on the oldest initial wealth-

holders which, in combination with its other efficiency-enhancing provisions, allows efficiency gains

that benefit not only long-run generations, but even most transition generations. Second, the X tax

preserves progressivity, and so spreads these gains over those in different income groups.

36

However, even the reform with the worst long-run performance, the flat tax with transition

relief, with a low value of , is favored by over two-thirds of the initial population, as the

combination of transition relief and adjustment costs makes older individuals better off, and low-end

progressivity helps those with lower lifetime income. As this simulated reform reduces long-run

output and welfare for the vast majority of the population, the voting measure serves as a reminder

that the shape of tax reform will not necessarily reflect concern about the economy’s long-run

condition.

VIII. Summary and Conclusion

Proponents and opponents of fundamental tax reform wrestle with the same question: Are

the gains to the winners worth the costs to the losers? The answer involves value judgements that

go beyond economic science. But forming one’s judgements requires knowing what fundamental

tax reform will do to the economy and to its current and future participants. This paper tries to

provide a better sense of those outcomes by simulating fundamental tax reform in a much-improved

version of the Auerbach-Kotlikoff model—one that considers intra- as well as intergenerational

equity and one that is closely calibrated to U.S. fiscal institutions and tax policy.

The model predicts significant long-run increases in output from replacing the current U.S.

federal tax system with a proportional consumption tax. For our base case, output would rise

eventually by more than 9 percent. For middle- and upper-income classes alive in the long run, this

policy is a big winner. But older transition generations suffer from the imposition of an implicit

capital levy, and low-income individuals, even in the long run, suffer a significant loss as growth

fails to compensate for the decline in tax progressivity.

37

The flat tax, which modifies the basic consumption tax by exempting housing wealth from

taxation and by providing a large wage-tax deduction, improves the welfare of lower-income

individuals in the long run, but at a cost of more than halving the economy’s long-run output rise,

to 4.5 percent for the same economic assumptions. Even then, this reform leaves initial older

generations worse off. Insulating them through transition relief, in the form of maintaining present-

law depreciation allowances on existing capital, further reduces the long-run output increase — to

just 1.9 percent.

Other reforms produce similar tradeoffs. Switching to a proportional income tax hurts

current and future low lifetime earners but helps everyone else. The X tax, which combines

consumption-tax and progressive wage-tax elements, makes everyone better off in the long run and

raises output by even more than the flat tax. But this reform harms initial older generations who face

an implicit tax on their wealth. Further, with smaller but still plausible labor supply and saving

responses to tax changes, the potential gains and hence the scope for trading off efficiency for equity

are reduced. Indeed, for a low intertemporal elasticity of substitution, a flat tax that offers transition

relief actually reduces output in the long run — presumably not what proponents of tax reform have

in mind.

Thus, the view formed on the basis of our model is that fundamental reform of the current

U.S. tax structure offers the possibility of significant macroeconomic expansion and welfare gains

for many, but not without true sacrifice by certain groups. Adjustments that attempt to prevent

adverse distributional effects yield much more modest aggregate effects. While we have not sought

to identify more complicated policies that shield all losers from sacrifice, our findings do suggest

38

that such policies, if they exist, are likely to yield much output increases than those of policies that

provide no such relief.

While we have adapted our model to accommodate many of the key issues that arise when

considering tax reform, there are some that we have not addressed. In treating the economy as

closed, we may have overstated the depressing impact of additional capital accumulation on the rate

of return and, hence, understated the potential welfare and output gains from tax reform, a conclusion

consistent with the open-economy simulations presented by Auerbach (1996).

Additional gains may also accrue under any of the tax reforms considered here as a result of

the more uniform treatment of different types of capital and different sectors of production. There

may also be welfare gains from tax simplification arising through reduced costs of compliance and

enforcement. Gains from both of these sources, though, would depend on the reformed tax system

maintaining not only its reduced rates and broader base, but also it lack of special interest provisions

that exist under the current tax system and could arise under at least some of the alternatives we have

considered. Finally, as discussed recently by Gentry and Hubbard (1997) the distributional and

efficiency effects of tax reform might change somewhat in a richer economic model incorporating

variation in risk and rates of return, for one would then need to consider the risk-sharing aspects of

different tax systems and the extent to which the capital levy of the consumption tax hit not only the

quasi-rents of existing capital, but also true economic rents. A challenging task for future research

is to determine the relative importance of these factors.

39

References

Altig, David, and Charles T. Carlstrom, “Marginal Tax Rates and Income Inequality in a Life-CycleModel,” American Economic Review, forthcoming.

Auerbach, Alan J., “Tax Reform and Adjustment Costs: The Impact on Investment and MarketValue,” International Economic Review, November 1989, 939-62.

Auerbach, Alan J., “Tax Reform, Capital Accumulation, Efficiency, and Growth,” in Henry J. Aaronand William G. Gale, eds., Economic Effects of Fundamental Tax Reform, The BrookingsInstitution: Washington, 1996, 29-81.

Auerbach, Alan J., Jagadeesh Gokhale, Laurence J. Kotlikoff, John Sabelhaus, and David N. Weil,“The Annuitization of Americans’ Resources: A Cohort Analysis” NBER Working Paper No. 5089,April 1995.

Auerbach, Alan J. and Laurence J. Kotlikoff, Dynamic Fiscal Policy, Cambridge University Press:Cambridge, 1987.

Auerbach, Alan J., Laurence J. Kotlikoff, Robert P. Hagemann, and Giuseppe Nicoletti, “TheEconomic Dynamics of an Ageing Population: The Case of Four OECD Countries,” OECDEconomic Studies, Spring 1989, 51-96.

Auerbach, Alan J. and Joel Slemrod, “The Economic Effects of the Tax Reform Act of 1986,”Journal of Economic Literature, June 1997, 589-632.

Blundell, Richard, Costas Meghir, and Pedro Neves, “Labour Supply and IntertemporalSubstitution,” Journal of Econometrics, September 1993, 137-60.

Bradford, David, Untangling the Income Tax, Harvard University Press: Cambridge, 1986.

Browning, Martin, Lars Peter Hansen, and James J. Heckman, “Micro Data and General EquilibriumModels,” unpublished manuscript, September 1998.

Congressional Budget Office, Comparing Income and Consumption Tax Bases, CBO Paper, July1997.

Cummins, Jason G., Kevin A. Hassett, and R. Glenn Hubbard, “A Reconsideration of InvestmentBehavior Using Tax Reforms as Natural Experiments,” Brookings Papers on Economic Activity,Spring 1994, 1-74.

Cutler, David M., “Tax Reform and the Stock Market: An Asset Price Approach,” AmericanEconomic Review, December 1988, 1107-17.

40

Engen, Eric M. and William G. Gale, “The Effects of Fundamental Tax Reform on Saving,” inHenry J. Aaron and William G. Gale, eds., Economic Effects of Fundamental Tax Reform, TheBrookings Institution: Washington, D.C., 1996, 83-121.

Fullerton, Don and Diane Lim Rogers, Who Bears the Lifetime Tax Burden? The BrookingsInstitution: Washington, 1993.

Gentry, William M., and R. Glenn Hubbard, “Distributional Implications of Introducing a Broad-Based Consumption Tax,” in J. Poterba, ed., Tax Policy and the Economy 11, MIT Press:Cambridge, 1997, 1-47.

Hall, Robert E. and Alvin Rabushka, The Flat Tax, Hoover Institution Press: Stanford, 1983; 2nded., Hoover Institution Press: Stanford, 1995.

Hubbard, R. Glenn and Kenneth Judd, “Social Security and Individual Welfare: PrecautionarySaving, Borrowing Constraints, and the Payroll Tax,” The American Economic Review, September1987, 630-46. Hubbard, R. Glenn, Jonathan Skinner, and Stephen P. Zeldes, “Precautionary Saving and SocialInsurance,” Journal of Political Economy, April 1995, 360-99.

Kotlikoff, Laurence J., and Lawrence H. Summers, “The Role of Intergenerational Transfers inAggregate Capital Accumulation,” Journal of Political Economy, August 1981, 706-32.

MaCurdy, Thomas E., “An Empirical Model of Labor Supply in a Life-Cycle Setting,” Journal ofPolitical Economy, December 1981, 1059-85.

Menchik, Paul L. and Martin David, “The Incidence of a Lifetime Consumption Tax.” National TaxJournal, 35, 1982, 189-203.

Mulligan, Casey B., “Substitution Over Time: Another Look at Life-Cycle Labor Supply,” in BenS. Bernanke and Julio J. Rotemberg, eds., NBER Macroeconomics Annual 1998, MIT Press:Cambridge, 1998, 75-134.

Slemrod, Joel and Jon Bakija, Taxing Ourselves: A Citizen’s Guide to the Great Debate over TaxReform, MIT Press: Cambridge, 1996.

Stokey, Nancy L. and Segio Rebelo, “Growth Effects of Flat-Rate Taxes,” Journal of PoliticalEconomy, June 1995, 519-50.

Ziliak, James P. and Thomas J. Kniesner, “Estimating Life Cycle Labor Supply Effects,” Journalof Political Economy, April 1999, 326-59.

25Our use of the variable 8 here follows the notation found in the relevant literature, and should not be confusedwith its use in the rest of the paper, to represent the rate of technological progress.

41

lt'(wt )& (1% w 1&

t )&

1& x( )(A1)

'(lL

)( % (1& ))(A2)

Appendix

Labor Supply Elasticities

This section of the appendix discusses the relationship between parameters of the utility

function given in expression (1) in the text and those typically estimated in the labor supply

literature. Papers in this literature contain a variety of different labor supply elasticity concepts.

However, perhaps most useful from our perspective is that of the “ -constant” or Frisch elasticity

of labor supply (e.g., MaCurdy 1981), which measures the variation in labor supply along an optimal

path with the marginal utility of income constant.25 In some studies, elasticities estimated in this

way are also referred to as an intertemporal elasticities of substitution, but this equivalence does not

hold for our specification of preferences, which are not separable between goods and leisure.

For our time-separable utility function, a 8-constant change in the after-tax wage, wt, affects

only consumption and leisure at date t. Thus, given the optimal path for these variables (Auerbach

and Kotlikoff 1987, p. 31, expressions 3.11 and 3.12), date-t leisure may be shown to satisfy:

where x(8) does not depend on wt. Using (A1), we derive the following expression for the 8-

constant elasticity of labor supply, Lt, with respect to wt:

where

26Other recent papers in the literature include Blundell, Meghir and Neves (1993), Mulligan (1998), and Ziliakand Kniesner (1999).

42

'w 1&

t

1% w 1&t

(A3)

Note that . corresponds to leisure’s “share” in the within-period utility function. Since we calibrate

the model for different values of D and ( so that this share is roughly .6 (and the consumption/labor

share is .4), the value of 0 from (A2) is roughly 1.5(.6( + .4D). For our base case values of D=.8 and

(=.25, this gives a value of 0=.7. This elasticity is reasonable, given the range of values estimated

in the literature, some of which are surveyed in Browning, Hansen and Heckman (1998).26

Estimates for men are in some cases higher, but typically somewhat lower, while estimates for

women are generally at least as high, and in some cases much higher. Our alternative simulations

that use values of D=.4 and (=.10, yield corresponding values for 0 of .46 and .57, respectively.

Calculation of Earnings-Ability Profiles

Our earnings-ability profiles are based on the individual files of the University of Michigan’s

Panel Study of Income Dynamics (PSID) from 1976-1988. These calculations generally follow those

of Fullerton and Rogers (1993, Chapter 4), except where indicated.

The sample utilized contains 9335 observations on 891 individuals. We excluded individuals

with imputed real hourly earnings less than one dollar and those with clear inconsistencies in levels

of educational attainment over the time period. This sample differs from that used by Fullerton and

Rogers, who aggregate individual observations into household observations and consider only

households with stable marital histories. We used the following procedure to obtain wage profiles.

27This difference is minor, as actual wages are observed for only a small portion of any given individual’sworking life. As the explanatory variables include individual fixed effects, 0i, and age interacted with other individualcharacteristics, the use of fitted rather than actual values during the period of observed wages amounts to smoothingout high-frequency wage fluctuations.

43

j3h'1 h%5AGEitDUM h

it % 9AGE 2it EDUit % j3

h'1 h%9AGE 2it DUM h

it % it

wit ' 0i % 1BYi % 2AGEit % 3AGE 2it % 4AGE 3

it % 5AGEitEDUit %

(i) First stage regressions were run on the entire data set using a common set of explanatory

variables. The specification is identical to that used by Fullerton and Rogers except a birth-year

variable, which is added to control for age-cohort effects. It is given by

where

Ðit = the log of the real hourly wage for person i in time t,

0i = fixed effect for person i,

BYi = the birth year of person i,

AGEit = the age of person i at time t,

EDUit = the education level (in years) of person i in time t,

DUMith = dummy variables for the marital status (h=1), race (h=2), and sex (h=3)of

person i in time t; potentially variable over time only for h=3.

Table A1 reports the resulting coefficient estimates.

(ii) The coefficient estimates obtained from step (i) were used to construct simulated

life-cycle wage profiles for each individual from age 21 through 80. Unlike Fullerton and Rogers,

we do not splice wage observations from the PSID with predicted values in generating the profiles.

Instead we simply use predicted values for all wage observations.27 In constructing these profiles

44

LIi ' j80s'21 (1%r)&(s&21) (wis@ 4000)

w jis '

j0 %

j1AGE j

is % 2(AGE jis)

2 %j3(AGE j

is)3 %

jis.

we set education to the highest reported level and assume that marital status is constant and equal

to married if the individual is married at any time over the PSID sample period.

(iii) Lifetime wage income (LI) levels are imputed from the profiles generated in step (ii)

according to the formula

where the discount rate r is set to 8 percent, Ðis is the predicted wage of the individual at age s, and

4000 is the potential full-time endowment of work hours. This calculation follows Fullerton and

Rogers, except for the choice of r.

All observations were next sorted in ascending order of lifetime wage income, and divided

into the twelve groups described in the text. Following the text and letting j be the fraction of the

population in group j, our division yields j = 0.02 for j = 1 and 12, j =0.08 for j =2 and 11, and

j = 0.1 for j = 3 through 10.

(iv) Finally, regressions of the predicted wage observations on a common group intercept and

a cubic in age were run for each group:

The estimated coefficients appear in Table A2. Figure 1’s profiles are based on these estimates

adjusted for annual efficiency growth of one percent. Taking account of the various differences in

methodology, they exhibit a pattern similar to those in Fullerton and Rogers.

Table 1. Benchmark Parameter Definitions and Values

Symbol Definition Value

PreferencesUtility weight on leisure 1.00Rate of time preference 0.004Intertemporal substitution elasticity 0.25

µj Utility weight placed on bequests by income class j1

Intratemporal substitution elasticity 0.80

Human CapitalProductivity of agent in income class j at age s.2j

s

Demographicsn Population growth 0.01N Number of children per adult, (1+n)20 1.22

j Fraction of agents of income class j3

TechnologyRate of technological change 0.01Adjustment cost parameter 0.10Net capital share 0.25Constant elasticity of substitution 1.00

Values of Fiscal Variables in Initial Steady State— Debt service as fraction of National Income 0.0310— Disability Insurance tax rate 0.0190— Medicare (HI) tax rate 0.0290— Social Security (OASI) tax rate 0.0990— Social Security replacement rate4

— Social Security marginal tax-benefit linkage 0.25— Payroll tax ceiling $62,700

C Proportional consumption tax 0.113K Proportional capital income tax 0.20

W(@) Progressive wage tax with deductions & exemptions5 Y State proportional income tax less evasion adjustment 0.011

— Reduction of wage base from itemized deductions6 0.0755— Reduction of wage base from fringe benefits6 0.1129z Expensing fraction7 0.20

1 Calibrated in the initial state to match the level of bequests—as a fraction of mean national income—inFullerton and Rogers (1993, Table 3-8), in 1996 dollars.2 See Appendix for estimation procedure.3 1=0.02, 2=0.08, i=0.10 (3#i#10), 11=0.08, 12=0.024 The statutory progressive bendpoint formula for 1996, scaled up by a factor of 2 to account for the fact thatother non-DI benefits (mainly spousal and survivors benefits) account for 50% of all benefits paid (see 1996OASDI Trustees Report, Table II.C7).5 The 1996 statutory tax function for a single individual with a deduction equal to $9661 ($4,000 standarddeduction, $2,550 personal exemption and $2,550@N exemption for dependents).6 Total proportional base reduction above the standard deduction therefore equals 0.18845.7 Deductions for new investment above economic depreciation and adjustment costs.

Table 2. Key Variables in the Initial Steady-State and U.S. Data

Model Empirical Estimate and Calculation

Concept Value Estimate Calculation (using NIPA unless indicated)

Composition of National Income (fraction)

Personal Consumption 0.731 0.720 Personal consumption expenditures - housing services

Net Saving Rate 0.051 0.056 (National saving - capital consumption allowance)/NI

Government Purchases 0.211 0.212 Consumption expenditures + gross investment forfederal (defense and nondefense) and state and local -consumption of fixed capital

Tax Rates and Revenue

Avg. Marginal Wage Tax1 0.216 0.217 Auerbach (1996) based on the NBER TAXSIM model.

Government Revenue 0.239 0.239 Total receipts - contributions for social insurance -property taxes (state and local)

OASDHI Tax 0.146 0.147 1996 tax rate is 15.3 which includes trust fundcontributions equal to about 0.6.

The Capital-Output Ratio and the Pre-Tax Rate of Return

Capital-Output Ratio 2.562 2.660 (1993 current-cost net stock of fixed reproducible wealth- gov’t owned fixed capital) / 1993 National Income

Pre-Tax Rate of Return2 0.083 0.093 The 1960-1994 average of the sum of interest, dividends,retained earnings and all corporate taxes divided by thereplacement value of capital stock (Rippe, 1995).

1 Does not include the payroll tax.2 The social marginal rate of return (i.e., before corporate taxes).

Table 3. Key Elements of Tax Reform Experiments

Experiment Description

Proportional Income Tax Eliminate all Tax Base ReductionsEliminate the standard deduction, personal exemption,exemptions for dependents, itemized deductions, preferentialtax treatment of all fringe benefits (the consumption taxtreatment of pension and the deductibility of non-pensionbenefits), and the deductibility of state income taxes at thefederal level.1

Flattening of Tax RatesReplace progressive wage tax and proportional capital incometax with a proportional equal tax rate on wage and capitalincome. Eliminate double taxation of capital income.

Proportional Consumption Tax Eliminate all Tax Base ReductionsFlattening of Tax RatesFull Expensing

Allow the deductibility of all new investment.

Standard Flat Tax Eliminate all Tax Base ReductionsFlattening of Tax RatesFull ExpensingProtection of Housing Wealth

Housing (including consumer durables) remain untaxed.2

Standard DeductionAllow for a deduction for a single individual equal to $9,500.

Flat Tax with Transition Relief Eliminate all Tax Base ReductionsFlattening of Tax RatesFull ExpensingProtection of Housing WealthStandard DeductionTransition Relief

All existing assets continue to receive depreciationallowances.3

X Tax Eliminate all Tax Base ReductionsPreserve Current-Law Progressive Wage Tax4

Capital Income Tax Set at Highest Marginal Wage Tax RateFull ExpensingProtection of Housing Wealth5

1 Consumption tax treatment of pensions is eliminated by decreasing the consumption tax by 0.025 and subjecting all compensationto the new proportional income tax.2 About 50% of the capital stock is composed of housing and consumer durables whose imputed rent is not taxed. Hence, theproportional tax rate on capital income is set to half of the tax rate on wage income.3 As noted in Auerbach (1996, footnote 46), under current-law and with current inflation, the present value of remainingdepreciation allowances per dollar of net nonresidential capital is approximately half the value of the assets. Allowing for thesedepreciation allowances has the same impact as forgiving half of the cash-flow tax on existing assets. Hence, the cash flow tax oncapital income is set to one quarter of the replacement proportional wage tax rate.4 General equilibrium effects and the constant government revenue constraint requires proportional shifts in the wage tax schedule(with an increase in the short run and a decrease in the long run). The average marginal tax rate is reported in Table 4.5 Since the highest marginal wage tax rate in the final steady state equals about 0.30, the capital income tax is set equal to 0.15.

Table 4. Base Case Results, Five Tax Reforms

YearNationalIncome1

CapitalStock1

LaborSupply1

NetSavingRate

Before-TaxWage1

After-TaxWage,2

InterestRate

Normal-izedTobin’s q

TaxRate3

PayrollTax Rate

1996 1.000 1.000 1.000 0.051 1.000 0.775 0.083 1.000 0.216 0.146

Proportional Income Tax 1997 1.038 1.002 1.051 0.056 0.988 0.825 0.088 1.037 0.135 0.142

2010 1.044 1.030 1.050 0.054 0.995 0.835 0.083 1.028 0.131 0.142

2145 1.049 1.056 1.047 0.052 1.001 0.842 0.083 1.019 0.130 0.144

Proportional Consumption Tax 1997 1.044 1.010 1.063 0.073 0.987 0.817 0.079 0.960 0.142 0.140

2010 1.063 1.108 1.054 0.067 1.013 0.843 0.076 0.934 0.138 0.140

2145 1.094 1.254 1.046 0.059 1.046 0.881 0.073 0.906 0.127 0.141

Flat Tax (Standard) 1997 1.010 1.006 1.016 0.065 0.997 0.754 0.076 0.964 0.214 0.145

2010 1.022 1.059 1.013 0.061 1.011 0.767 0.078 0.958 0.211 0.145

2145 1.045 1.150 1.013 0.056 1.032 0.796 0.080 0.941 0.199 0.144

Flat Tax (Transition Relief) 1997 0.995 1.003 0.994 0.059 1.002 0.730 0.081 1.001 0.241 0.148

2010 1.005 1.031 0.998 0.057 1.008 0.741 0.080 0.994 0.234 0.146

2145 1.019 1.083 0.998 0.055 1.021 0.759 0.078 0.983 0.226 0.146

X Tax 1997 1.018 1.009 1.027 0.069 0.996 0.789 0.063 0.949 0.178 0.144

2010 1.031 1.076 1.019 0.064 1.014 0.804 0.077 0.910 0.177 0.144

2145 1.064 1.210 1.020 0.059 1.044 0.849 0.074 0.882 0.157 0.143

1 Measured per-effective labor unit and indexed with a value of 1.00 in 1996.2 Computed as (1- )@(before-tax wage), where is the economy-wide effective average marginal tax rate on wage income, taking state income taxes and tax evasion into account.3 Statutory federal rate; for 1996, this is the rate that applies to wage income.

Table 5. Sensitivity Analysis: Production Parameters

YearNationalIncome1

CapitalStock1

LaborSupply1

NetSavingRate

Before-TaxWage1

After-TaxWage2

InterestRate

Normal-izedTobin’s q

TaxRate3

PayrollTax Rate

= 0

1996 1.000 1.000 1.000 0.051 1.000 0.775 0.097 1.000 0.214 0.147

Flat Tax (Standard) 1997 1.027 1.019 1.030 0.087 0.997 0.754 0.098 0.901 0.214 0.144

2010 1.044 1.135 1.015 0.070 1.028 0.785 0.090 0.905 0.207 0.143

2145 1.059 1.215 1.012 0.059 1.047 0.813 0.085 0.911 0.194 0.143

Flat Tax (Transition Relief) 1997 1.006 1.014 1.003 0.075 1.003 0.731 0.097 0.950 0.242 0.147

2010 1.022 1.091 1.000 0.064 1.022 0.755 0.091 0.952 0.231 0.145

2145 1.035 1.153 0.998 0.057 1.037 0.778 0.087 0.955 0.220 0.145

= .8

1996 1.000 1.000 1.000 0.051 1.000 0.775 0.083 1.000 0.216 0.146

Flat Tax (Standard) 1997 1.009 1.006 1.014 0.064 0.994 0.751 0.077 0.967 0.215 0.146

2010 1.022 1.055 1.013 0.060 1.010 0.767 0.078 0.952 0.210 0.144

2145 1.043 1.136 1.015 0.056 1.032 0.797 0.078 0.936 0.198 0.143

Flat Tax (Transition Relief) 1997 0.994 1.003 0.993 0.059 1.000 0.729 0.081 1.000 0.241 0.148

2010 1.004 1.029 0.998 0.057 1.007 0.741 0.080 0.993 0.234 0.146

2145 1.018 1.075 1.000 0.054 1.020 0.759 0.081 0.982 0.226 0.145

1 Measured per-effective labor unit and indexed with a value of 1.00 in 1996.2 Computed as (1- )@(before-tax wage), where is the economy-wide effective average marginal tax rate on wage income, taking state income taxes and tax evasion into account.

3 Statutory federal rate; for 1996, this is the rate that applies to wage income.

Table 6. Sensitivity Analysis: Preference Parameters

YearNationalIncome1

CapitalStock1

LaborSupply1

NetSavingRate

Before-TaxWage1

After-TaxWage2

InterestRate

Normal-izedTobin’s q

TaxRate3

PayrollTax Rate

=.4

1996 1.000 1.000 1.000 0.049 1.000 0.775 0.086 1.000 0.213 0.141

Flat Tax (Standard) 1997 1.008 1.006 1.013 0.063 0.998 0.756 0.081 0.971 0.213 0.140

2010 1.016 1.057 1.005 0.059 1.013 0.769 0.081 0.953 0.211 0.139

2145 1.028 1.136 0.996 0.055 1.033 0.793 0.078 0.935 0.203 0.139

Flat Tax (Transition Relief) 1997 1.001 1.003 1.002 0.057 1.000 0.735 0.084 1.005 0.235 0.141

2010 1.007 1.033 1.000 0.055 1.008 0.746 0.083 0.995 0.227 0.140

2145 1.015 1.080 0.995 0.054 1.021 0.760 0.082 0.987 0.225 0.140

=.1

1996 1.000 1.000 1.000 0.051 1.000 0.775 0.083 1.000 0.216 0.146

Flat Tax (Standard) 1997 1.007 1.003 1.010 0.057 0.998 0.757 0.075 0.947 0.211 0.146

2010 1.007 1.019 1.004 0.054 1.004 0.756 0.082 0.932 0.217 0.146

2145 1.013 1.038 1.005 0.053 1.008 0.762 0.081 0.928 0.213 0.146

Flat Tax (Transition Relief) 1997 0.990 1.000 0.986 .050 1.003 0.734 0.077 0.974 0.239 0.147

2010 0.985 0.985 0.984 .047 1.000 0.726 0.083 0.965 0.244 0.149

2145 0.969 0.920 0.983 .049 0.983 0.702 0.088 0.975 0.256 0.149

1 Measured per-effective labor unit and indexed with a value of 1.00 in 1996.

2 Computed as (1- )@(before-tax wage), where is the economy-wide effective average marginal tax rate on wage income, takingstate income taxes and tax evasion into account.3 Statutory federal rate; for 1996, this is the rate that applies to wage income.

Table 7. Percentage of Initial Population with Gains from Reform

Proportional Income Tax 74.0

Proportional Consumption Tax 66.8

Flat Tax 73.5

Flat Tax with Relief 68.2

X Tax 84.7

Flat Tax, no adjustment costs 74.1

Flat Tax with Relief, no adjustment costs 68.0

Flat Tax, low 71.6

Flat Tax with Relief, low 68.0

Flat Tax, low 66.2

Flat Tax with Relief, low 64.2

Flat Tax, low 70.8

Flat Tax with Relief, low 68.4

Table A1. PSID Regression Results

Variable Coefficient T statistic

Birth Year (BY) -0.0005 1.47

Age (AGE) 0.0883 3.46

Age-Squared -0.0016 3.20

Age-Cubed 6.66E-6 2.07

Age x Education (EDU) -0.0009 2.02

Age x Marital Status (DUM1) 0.0081 3.37

Age x Race (DUM2) 0.0119 2.81

Age x Sex (DUM3) 0.0339 1.69

Age-Squared x Education 2.16E-5 3.30

Age-Squared x Marital Status -0.0001 2.84

Age-Squared x Race -0.0002 2.70

Age-Squared x Sex -0.0004 2.00

Adjusted R2 0.9779 —

Table A2. Estimated Wage Profile Coefficients by Lifetime Income Group

LI Group Intercept Age Age-Squared Age-Cubed

1 -0.6421 0.0949 -0.00158 7E-06

2 -0.2294 0.0941 -0.00157 7E-06

3 0.1831 0.0929 -0.00156 7E-06

4 0.4693 0.0907 -0.00155 7E-06

5 0.6772 0.0882 -0.00150 7E-06

6 0.8865 0.0853 -0.00147 7E-06

7 0.9794 0.0884 -0.00151 7E-06

8 1.1606 0.0864 -0.00148 7E-06

9 1.3180 0.0855 -0.00147 7E-06

10 1.4814 0.0862 -0.00147 7E-06

11 1.8151 0.0856 -0.00146 7E-06

12 2.5745 0.0853 -0.00146 7E-06